Research map / Large language models / Retrieval and knowledge
ARTEM: Enhancing Large Language Model Agents with Spatial-Temporal Episodic Memory
Cassandra Hui-Ming Tan, Budhitama Subagdja, Ah-Hwee Tan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use LLM-based answer synthesis on top of STEM (ARTEM vs STEM) | STEM · single-event queries (Bin 1) | F1: 0.55 -> 0.68 | Improved |
| add LLM synthesis to STEM for latest-event queries | STEM · Latest time instruction | F1: 0.54 -> 0.66 | Improved |
| add LLM synthesis to STEM for full-event-detail cues | STEM · Full event details cue | F1: 0.5 -> 0.67 | Improved |
| replace STEM retrieval with LLM synthesis on null queries | STEM · 0 matching events | F1: 1.00 -> 0.95 | Hurt |
| replace STEM with ARTEM on chronological recall | STEM · chronological task | F1: 0.644 -> 0.585 | Hurt |
| replace STEM with ARTEM on simple recall | STEM · simple recall | F1: 0.733 -> 0.709 | Hurt |
| use STEM episodic memory instead of in-context gemini-2-pro for chronological recall | gemini-2-pro · chronological task | F1: 0.290 -> 0.644 | Improved |
| use reasoning-optimized model o1 in-context instead of gpt-4o | gpt-4o · chronological task | F1: 0.204 -> 0.052 | Hurt |
| fine-tune gpt-4o-mini instead of in-context for null queries | gpt-4o-mini · 0 matching events | F1: 0.51 -> 0.00 | Hurt |
| fine-tune gpt-4o-mini instead of in-context for single-event queries | gpt-4o-mini · 1 matching event | F1: 0.54 -> 0.83 | Improved |
| fine-tune gpt-4o-mini instead of in-context for 6+ events | gpt-4o-mini · 6+ matching events | F1: 0.50 -> 0.19 | Hurt |
| add RAG to gpt-4o-mini instead of in-context | gpt-4o-mini · 6+ matching events | F1: 0.50 -> 0.62 | Improved |
Closest papers by meaning
- Episodic Memories Generation and Evaluation Benchmark for Large Language Models (ICLR 2025 · similarity 0.92)
- REMem: Reasoning with Episodic Memory in Language Agent (ICLR 2026 · similarity 0.91)
- Human-inspired Episodic Memory for Infinite Context LLMs (ICLR 2025 · similarity 0.91)
- A-Mem: Agentic Memory for LLM Agents (NeurIPS 2025 · similarity 0.91)
- Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions (ICLR 2026 · similarity 0.91)
- AMA-Bench: Evaluating Long-Horizon Memory for Agentic Applications (ICML 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
